The individual blocks on the SD card have a limited life cycle, like 1000000 writes or something like that. Once those writes are exhausted, they can longer be erased and written to. So if you created a swapfile and that swapfile occupies the same physical area on the card all the times, that area will wear out quickly if linux is constantly swapping inactive RAM data from/to that area. I know we use the SD card to save data / apps / etc, but writing is limited on that (unless you're constantly installing software every second of the day); also, the NAND, card driver under WinMo and Linux is smart because they're designed to do wearleveling; that is, spread out the writes evenly all over the area of the Nand/card so that no one area gets exhausted quickly. On a SD card or NAND, I believe each block contains a header section that tells the OS how many writes have been written to it, so this is how it knows what blocks have been written to more or less; I'm not sure but I believe that's how it works. I believe "reading" is unlimited but if you can no longer write to a worn out SD card, it's useless. So that's why using swapping is not recommended. The advanced features of swapper 2 is nice; you can tell it to delete the swapfile and create a new one each time android boots; when creating the new swapfile, it does at a different area of the Sd card so the previous area is not exhausted out. I believe hard drives have a much longer write life cycle. Any expert, please correct me if I'm wrong on this.

Edit: oops, everyone beats me to the explanation :oops: And mine is inaccurate about the write cycles, I guess it's in the thousands, not millions, which means you can easily wear out that card using swapping.

Sent from my FRX05-ized TP2

I'm not sure about Sd cards, but I know that on the Nand, the individual unit of storage is a block. Depending the the Nand hardware, a block can be 32kb, 64kb, 128kb, etc. The write cycle is per block. So if you copy a file, let's say 64kb and the block can hold 32kb, the nand driver will use 2 blocks (plus any overhead) to write that file; the driver is smart enough to use wearleveling to spread out the writes over the whole Nand; it may be similarly with a sd card.

Sent from my FRX05-ized TP2

Re: [TESTING] Updated 3/11/11 New Kernel - TONS of changes!!
 

thanks guys---a lot of good info.

personal experience. facebook widget (the big one) tends to cause locks/no wakeups on the new kernel

swappy2 - a marked improvement, I have since changed my settings due to the great info from ya'll (plus I was looking at the info page and wasn't caching as much as I thought I would)

dropped my settings to
swap size - 64MB (down from 128)
swappiness - 20 (down from 60)
recreate swap file - checked
reformat swap - unchecked

speed is still amazing!

Re: [TESTING] Updated 3/11/11 New Kernel - TONS of changes!!
 

ffkip911, here's a forum on swapper 2 in which some folks used it on the Cyanogen android phones:

Swapper 2 - CyanogenMod Forum. Some good info from users.

If you have a very slow SD card, I don't think using swapper 2 is recommended as it will be very slow. But then again, there are known problems with fast SD cards (class 6 and greater).

Re: [TESTING] Updated 3/11/11 New Kernel - TONS of changes!!
 

well, if flash can only be written onto a specific number of times----what happens to internal memory on the phone---every time I get a call, it is logged (stored) onto internal memory. at the end of the day, i erase it, only to get more calls that are logged (stored) at the same location on the memory....to follow the theory to its conclusion, I( should only get X number of days that the phone can work correctly -- ie log sms, mms, email, phone calls. after that, the internal memory would have met it's max number of write's and no longer store/log calls, sms, emails etc....

EDIT: I honestly have no idea how a flash drive actually stores information....playing devils advocate helps me understand. I am not discounting all the great info that has come out about swapping, I think this helps all involved to make the decision on whether to install the program or not.

Re: [TESTING] Updated 3/11/11 New Kernel - TONS of changes!!
 

Memory (RAM) doesn't work the same; it's volatile memory and never needs to retain the data, unlike SD card. Different hardware.

Random-access memory - Wikipedia, the free encyclopedia

LOL, I think you're a little obsessed with getting more "virtual" RAM out of your SD card ;). I wouldn't risk my SD card to be obliterated by using a swapfile; that's just a fact. This topic has probably been debated many times.

Re: [TESTING] Updated 3/11/11 New Kernel - TONS of changes!!
 

thanks---the topic has been discussed to it's conclusion. You guys rock

Re: [TESTING] Updated 3/11/11 New Kernel - TONS of changes!!
 

I've been using this kernel since Wednesday morning and wanted to add that it's been working really well for me.

I haven't had any "death sleep" issues, and the battery life is outstanding. There was one time where it froze when I pulled down the notification tray, but that's been the only issue so far. It's starting to feel like Android came on this phone from the factory.

Thanks to everyone that puts all the work in to make this possible!

Re: [TESTING] Updated 3/11/11 New Kernel - TONS of changes!!
 

Update: 3.11.11 kernel on freshly FAT32 formatted sd card, no OC, running well. No SODs, every day or two i get a reboot. Charging is still slow, but not running too hot. About 122F max temp, but I do live in SW FL...ambient is about 76-78F.

Re: [TESTING] Updated 3/11/11 New Kernel - TONS of changes!!
 

Good to hear!

Haven't had much in the way of kernels from WisTilt2, but I have been hashing out the new RIL with him.

Basically work on that is going great - but CDMA is really holding him up. He only has GSM devices, so he's kinda flying blind and depending on my (and rpierce99's) feedback.

So we're gonna keep chugging with that, progress is going well - I won't reveal too much, but there are going to be a lot of fixes when it is ready for general testing ;).